The Spatial Distribution of Chlorophyll in Leaves1[OPEN]

Aleca M. Borsuk and Craig R. Brodersen2,3

School of Forestry & Environmental Studies, Yale University, New Haven, Connecticut 06511

ORCID IDs: 0000-0002-1696-9647 (A.M.B.); 0000-0002-0924-2570 (C.R.B.).

Measuring and modeling the spatial distribution of chlorophyll within the leaf is critical for understanding the relationshipbetween leaf structure and carbon assimilation, for defining the relative investments in leaf tissues from the perspective of leafeconomics theory, and for the emerging application of in silico carbon assimilation models. Yet, spatially resolved leafchlorophyll distribution data are limited. Here, we used epi-illumination fluorescence microscopy to estimate relativechlorophyll concentration as a function of mesophyll depth for 57 plant taxa. Despite interspecific variation due todifferences in leaf thickness, mesophyll palisade fraction, and presence of large intercellular airspaces, the spatial distributionof chlorophyll in laminar leaves was remarkably well conserved across diverse lineages (ferns, cycads, conifers, ginkgo, basalangiosperms, magnoliids, monocots, and eudicots) and growth habits (tree, shrub, herbaceous, annual, perennial, evergreen,and deciduous). In the typical leaf, chlorophyll content increased gradually as a function of depth, peaking deep within themesophyll. This chlorophyll distribution pattern is likely coupled to adaxial and abaxial intraleaf light gradients, including therelative enrichment of green light in the lower leaf. Chlorophyll distribution for the typical leaf from our dataset was wellrepresented by a simple mathematical model (R2 = 0.94). We present chlorophyll distribution data and model equations formany ecologically and commercially relevant species and plant functional types (defined according to chlorophyll profilesimilarity, clade, and leaf thickness). These findings represent an advancement toward more accurate photosynthesismodeling and increase our understanding of first principles in intraleaf physiology.

The biochemistry of photosynthesis and biophysicalprocesses that constrain it are intrinsically linkedwithinthe landscape of the inner leaf. Because leaf tissue isheterogeneous in structure as well as photosyntheticcapacity (Smith et al., 1997; Adams and Terashima,2018; Earles et al., 2019), it follows that chlorophyllmay also be spatially heterogeneous. Yet, while bulkchlorophyll content of leaves can be readily measured(Porra et al., 1989; Palta, 1990), fewer studies havesystematically dissected the leaf to determine howchlorophyll content is partitioned across the photo-synthetic domain ormesophyll (Vogelmann and Evans,2002; Johnson et al., 2005; Evans and Vogelmann, 2006;Brodersen and Vogelmann, 2010; Slattery et al., 2016).

The paucity of spatially resolved chlorophyll distribu-tion data available at present is likely due to thehistorically labor- and time-intensive processes ofobtaining such data, requiring extraction of chlorophyllfrom thin paradermal sections (Bornman et al., 1991;Cui et al., 1991; Nishio et al., 1993) or counting chloro-plasts with a light microscope (e.g. Aalto and Juurola,2002). However, epi-illumination chlorophyll fluores-cence offers a simple and rapid method of measuringrelative chlorophyll distribution within cell layers of aleaf (Vogelmann and Evans, 2002). In this method, afresh leaf is cut transversely, placed under amicroscope(Fig. 1A), and irradiated with an excitation wavelengthorthogonal to the cut surface. Autofluorescence emittedfrom chlorophyll is captured with a charge-coupleddevice (CCD) camera with a bandpass filter to collectonly light generated from fluorescence. Pixel intensityin the grayscale images derived from this method isproportional to relative chlorophyll content accordingto a 1:1 relationship, as reported in detail by Vogelmannand Evans (2002). Relative chlorophyll content can thenbe measured directly from images using image analysissoftware (Fig. 1B) to obtain chlorophyll distribution(Fig. 1C). To the best of our knowledge, the epi-illumination fluorescence technique has only beenused to measure leaf chlorophyll distribution for sevenspecies—spinach (Spinacia oleracea; Vogelmann andEvans, 2002), Rhododendron catawbiense, Abies fraseri,Picea rubens (Johnson et al., 2005), Eucalyptus pau-ciflora (Evans and Vogelmann, 2006), snapdragon(Antirrhinum majus), and sunflower (Helianthusannuus; Brodersen and Vogelmann, 2010). Using lightsheet microscopy, a modified form of this method,

1This material is based upon work supported by the National Sci-ence Foundation Graduate Research Fellowship Program undergrant no. DGE1752134. Any opinions, findings, and conclusions orrecommendations expressed in this material are those of the author(s)and do not necessarily reflect the views of the National ScienceFoundation.

2Author for contact: craig.brodersen@yale.edu3Senior author.The author responsible for distribution of materials integral to the

findings presented in this article in accordance with the policy de-scribed in the Instructions for Authors (www.plantphysiol.org) is:Craig R. Brodersen (craig.brodersen@yale.edu).

C.R.B. conceived the original research plans and complementedthe experiments, data interpretation, and writing; A.M.B. performedmost of the experiments, analyzed the data, and wrote the manu-script with the contribution of C.R.B.; C.R.B. agrees to serve as theauthor responsible for contact.

[OPEN]Articles can be viewed without a subscription.www.plantphysiol.org/cgi/doi/10.1104/pp.19.00094

1406 Plant Physiology�, July 2019, Vol. 180, pp. 1406–1417, www.plantphysiol.org � 2019 American Society of Plant Biologists. All Rights Reserved. www.plantphysiol.orgon September 2, 2020 - Published by Downloaded from

Copyright © 2019 American Society of Plant Biologists. All rights reserved.

Slattery et al. (2016) were able to make similar mea-surements in a soybean (Glycine max) cultivar andmutant. These studies reported that chlorophyll contentin laminar leaves is nonuniform, with the lowest

concentration in the palisade mesophyll near the ad-axial surface and higher concentrations in the medialand abaxial domains. Johnson et al. (2005) introducedthe model equation

y ¼ y0 þ ax þ bx2

to express relative chlorophyll content (y) as a func-tion of leaf depth (x) in vertical profiles of R. catawbienseand A. fraseri (both laminar or relatively flattened)leaves. Yet, because of the lack of chlorophyll distri-bution data in the literature and the wide variability inleaf form, it is unclear whether this characterizationrepresents the rule or the exception.In theory, as a light-harvesting molecule, chlorophyll

should be well coordinated spatially with intraleaf lightavailability. Yet, Nishio et al. (1993) showed that carbonfixation gradients in S. oleracea leaves do not follow leafinternal light gradients, which decrease exponentiallythrough the leaf. Instead, carbon fixation was higher inthe middle of the leaf, with spongy mesophyll con-tributing significantly (40%) to total carbon reduction.This is largely counter-intuitive, assuming that carbonfixation should be highest at the upper leaf surface,where there is the most light available. Yet, observedincreases in chlorophyll within the middle of leaves hasbeen observed in multiple species (using paradermalsectioning; Bornman et al., 1991; Cui et al., 1991), withthe suggestion that this would balance the decline in theamount of available light from the adaxial and/or ab-axial surfaces. Here, we note that light gradients, andtherefore chlorophyll distribution, could be altered bypalisade mesophyll, which may propagate light deeperinto the leaf (Smith et al., 1997; Johnson et al., 2005).Studies have also demonstrated the importance of dif-ferent rates of light extinction through the leaf betweenstrongly absorbed (e.g. blue, red) and weakly absorbed(e.g. green) wavelengths. Cui et al. (1991) found thatwhile strongly absorbed wavelengths such as blue andred are 90% attenuated in the upper 20% of leaves,medial and deep leaf tissues remain relatively enrichedin green light. Interestingly, Sun et al. (1998) reportedthat green light drove more CO2 fixation than red orblue light in deep leaf tissues of S. oleracea. Evans andVogelmann (2003) also reported that photosyntheticcapacity matched the profile of green but not whitelight absorption in S. oleracea leaves. Taken together,studies such as these have contributed to the growingunderstanding of the role of green light in drivingphotosynthesis deep within leaf tissues (Terashimaet al., 2009; Smith et al., 2017). These reports are alsoconsistent with the concept of the intraleaf environmentas an analog to a canopy, with vertical gradients in lightquantity and quality. Terashima et al. (2009) noted,for example, that chloroplasts acclimate to the locallight environment, resulting in a gradient of sun andshade characteristics. Terashima et al. (2009) alsoreported that chloroplasts in the lowermost part ofS. oleracea leaves exhibit 20% to 40% of the maximal

Figure 1. Overview of the epi-illumination method. In brief, leaf sam-ples are dark-adapted for;20min before transverse sections (A) are cutand placed on damp filter paper for imaging. Samples are irradiated(,30 s), and fluorescence from chlorophyll is detected with a CCDcamera. Chlorophyll content is proportional to pixel intensity in theresulting grayscale images (B; Vogelmann and Evans, 2002). The 1Dchlorophyll distribution is measured in ImageJ using a line profile (linewidth mean of ;100 pixels, or ;60 mm at 203 magnification) drawnthrough the photosynthetic tissue. Nonphotosynthetic features such asveins (V), bundle sheath extensions (BSE), and upper epidermal (UE)and lower epidermal (LE) cells are excluded from the measurements.The profile begins at the upper edge of the mesophyll (e.g. palisademesophyll if present; PM) and ends at the lower edge of the spongymesophyll (SM). Previous studies using this technique included theepidermal cell layer (shown in C; blue bars), potentially making chlo-rophyll profiles look more bell-shaped than with this method. Chloro-phyll profiles are normalized as a function of maximum intensity toallow for comparison (absolute fluorescence intensity varies acrosssamples because of slight differences in exposure parameters and signaldecay over time) and are represented as a function of relative depth (C;0%, upper edge of mesophyll; 100%, lower edge of mesophyll).

Plant Physiol. Vol. 180, 2019 1407

Spatial Distribution of Chlorophyll

www.plantphysiol.orgon September 2, 2020 - Published by Downloaded from Copyright © 2019 American Society of Plant Biologists. All rights reserved.

photosynthetic capacity of the chloroplasts closest tothe adaxial surface. These data were used by Tholenet al. (2012) to model the fraction of chlorophyll inpalisade versus spongy mesophyll and the photosyn-thetic rate that could be sustained by green light. As-suming that S. oleracea chloroplasts in the lowermostpart of the leaf exhibit 20% of the maximal photosyn-thetic capacity of those in the uppermost part, photo-synthetic capacity was maximized by a roughly equal(;50%) distribution of chlorophyll between palisadeand spongy mesophyll tissues.

Based on these precedents, in a laminar leaf receivingirradiation primarily from the adaxial surface we an-ticipate that relative chlorophyll contentwould increasetoward the abaxial domain as a response to the declinein available light and/or the relative enrichment ofweakly absorbed green light and would be well char-acterized by the model equation y = y0 + ax + bx2 pro-posed by Johnson et al. (2005). Yet, carbon assimilationrequires more than light, namely CO2 and water. In-terestingly, leaf veins represent a nexus of distributionof these resources. Specifically, leaf vein-to-vein andvein-to-evaporative surface spacing is highly conserved(Noblin et al., 2008; Zwieniecki and Boyce, 2014), es-pecially in derived taxa. Thus, we also anticipated thatchlorophyll content might peak at a depth coincidentwith leaf veins.

The aim of this study was to describe the spatialdistribution of chlorophyll as a function of leaf depth inbulkmesophyll domains of laminar leaves across broadlevels of phylogenetic diversity. These observationswere intended to expand the number of species forwhich empirical chlorophyll distribution data areavailable, articulate any common scheme(s) in chloro-phyll distribution within bulk mesophyll domains, andinitiate an understanding of associations betweenchlorophyll distribution, light environment, leaf anat-omy, and the optimization of photosynthesis relative tothe multiple competing biophysical demands placed ona leaf.

Toward this, we used epi-illumination fluorescencemicroscopy to measure chlorophyll distributionswithin vertical (transverse) leaf profiles of 57 speciesfrom eight major terrestrial plant clades (ferns, cy-cads, conifers, ginkgo, basal angiosperms, magno-liids, monocots, and eudicots), representing adiversity of growth habit, habitat, and leaf form (e.g.tree, shrub, herbaceous, annual, perennial, ever-green, and deciduous plant species; SupplementalTable S1). Seven congeneric pairs (Acer, Arenga, Be-gonia, Dioon, Encephalartos, Quercus, and Zamia) wereincluded in this dataset to explore variability within agenus in basal and derived groups. Hierarchical clusteranalysis and mean model approaches were used to ar-ticulate patterns in chlorophyll distribution. We alsocharacterized leaf thickness, vein depth, and palisadefraction of mesophyll to examine possible relationshipsbetween patterns in chlorophyll content and major ana-tomical traits (Supplemental Table S1). In a subset of spe-cies,wemeasuredmonochromatic light absorptionprofiles

following themethods of Brodersen andVogelmann (2010)to characterize the relationship between intraleaf light en-vironment and chlorophyll distribution.

The results of this work could be used for researchfocused on the relative investments in leaf tissues fromthe perspective of leaf economics theory (Shipley et al.,2006; Reich, 2014) or the relationship between leafstructure and carbon assimilation (Terashima et al.,2011; Earles et al., 2019). This information is also rele-vant to the emerging application of in silico carbon as-similation models for testing first principles in plantphysiology and toward engineering increased perfor-mance and resiliency in agricultural crops (Tholen et al.,2012; Ort et al., 2015). While such models may incor-porate fine-scale phenomena such as intraleaf CO2diffusion (Pieruschka et al., 2006; Morison et al., 2007;Terashima et al., 2011; Tholen et al., 2014; Ho et al.,2016; Earles et al., 2018), hydraulics (Noblin et al.,2008; Zwieniecki and Boyce, 2014), and light propaga-tion (Smith et al., 1997, 2017; Sun et al., 1998;Vogelmann and Gorton, 2014; Ichiro et al., 2016), sim-ilarly resolved chlorophyll distribution data are not yetwidely available or well understood. This is evidentfrom the multiple approaches used for parameterizingchlorophyll distribution in leaf carbon assimilationmodels, including as a homogeneous biochemical do-main (Oguchi et al., 2011; Tholen and Zhu, 2011; Earleset al., 2017).

RESULTS

Chlorophyll Profiles of Individual Species

To visually assess the chlorophyll distribution, orchlorophyll profiles, of individual species, mean chlo-rophyll content was plotted as a function of meso-phyll depth (n = 3 profiles were collected for eachleaf, and n = 2–6 leaves were collected for eachspecies; Supplemental Fig. S1). The profile expressed byQuercus alba is representative of many of the species inthis study, where chlorophyll content was relativelyuniform or increased slightly toward the abaxial sur-face (Fig. 2, A–C). In some species, such as the fernPlatycerium bifurcatum (Fig. 2, D–F), chlorophyll contentwas clearly partitioned into two domains, concentrat-ing chlorophyll in either palisade and/or spongy me-sophyll just below the epidermis (Fig. 2E). Yet, for mostspecies, it was unclear whether mesophyll tissue typewas coordinated with changes in relative chlorophyllcontent (e.g. Dioon merolae, Amorphophallus titanum,Arenga engleri; Supplemental Fig. S1) or whether chlo-rophyll content was relatively uniform across the leaf,i.e. did not change despite differentiation of mesophyll(e.g. Q. alba, Fig. 2,A–C; Acer saccharum, Drimys winteri,Robinia pseudoacacia; Supplemental Fig. S1). The pres-ence of large intercellular airspaces depressed localchlorophyll content in some leaves, such as with theaquatic plant Eichhornia crassipes (Fig. 2, G–I), and thespatial heterogeneity of these intercellular airspaces led

1408 Plant Physiol. Vol. 180, 2019

Borsuk and Brodersen

www.plantphysiol.orgon September 2, 2020 - Published by Downloaded from Copyright © 2019 American Society of Plant Biologists. All rights reserved.

to increased variability compared to areas of the leafwith lower porosity (Fig. 2H). Where congeneric pairswere measured, the chlorophyll profiles were remark-ably similar, particularly in the cases of Acer andQuercus from the derived eudicot clade (SupplementalFig. S1).

Major Patterns in Chlorophyll Distribution

A hierarchical cluster analysis was used to identifycommon patterns in the chlorophyll profiles withoutimposing a priori classification. This indicated thepresence of six clusters or groups of species with similarchlorophyll profiles (Fig. 3A, a–f). To aid in comparingthe major patterns, the mean chlorophyll profile of eachgroup was analyzed according to the model equation(Eq. 1) using quadratic regression (Fig. 3B). Numericalcoefficients for the regression curves are found in Ta-ble 1. As discussed in detail in the “Materials andMethods” section, the first and last relative depthvalues (relative depth = 0, 100) or where mesophyllwas bordered by epidermal layers are not part of the

photosynthetic domain and are excluded from all re-gressions. Group (a) contained a single species, thebromeliad Billbergia minarum. This species had anatypical leaf form in our dataset due to a multiple-layerhypodermis (which was not included in measurementsbut may have influenced chlorophyll distribution in-directly). B. minarum had markedly lower chlorophyllcontent in the adaxial chlorenchyma (n = 1, R2 = 0.98,F(2,16) = 426.5, P, 0.000). Group (b) was dominated bygymnosperms (predominantly cycads; mean = 0.482mm, SE = 0.051). Species in group (b) exhibited a gradualincrease in chlorophyll from the adaxial side andmaxima shifted heavily to the abaxial domain (n = 11,R2 = 0.96, F(2,16) = 172, P , 0.000). Group (c) was thelargest group. This groupwas phylogenetically diverse,containing fern, cycad, magnoliid, monocot, and eudi-cot taxa and was typified by a uniform distribution(n = 22, R2 = 0.86, F(2,16) = 48.9, P , 0.000). Group (d)was also phylogenetically diverse, with a typical profilesimilar to groups (b) and (c), but with a gradual ta-pering of chlorophyll content at both adaxial and ab-axial edges (n = 15, R2 = 0.98, F(2,16) = 363.9, P, 0.000).The outlier E. crassipes comprised group (e). This was

Figure 2. Differences in chlorophyll distribution for three representative species. Q. alba (A–C), P. bifurcatum (D–F), andE. crassipes (G–I). At left, a typical grayscale fluorescence image is shown for each species (A, D, andG),where the yellow transectindicates the section of tissue in a single measurement; chlorenchyma domains such as palisade (PM) and spongymesophyll (SM)are included in the measurements, whereas nonphotosynthetic structures such as veins (V), upper epidermal (UE) and lowerepidermal (LE) layers, and hypodermal layers (HL) are excluded. Intercellular airspaces (IA) in E. crassipes are included in themeasurements and increase local variability in mean profiles. Heat maps (B, E, and H) of the same images help to highlight trendsin chlorophyll distribution. Multiple (n. 6) line profiles are measured (gray points; C, F, and I) to describe the mean chlorophyllprofile (black points; C, F, and I) for an individual species. Chlorophyll content (as a percentage of the absolute chlorophyllmaximum) is displayed as a function of relative depth (percentage depth from upper to lower edge of mesophyll).

Plant Physiol. Vol. 180, 2019 1409

Spatial Distribution of Chlorophyll

www.plantphysiol.orgon September 2, 2020 - Published by Downloaded from Copyright © 2019 American Society of Plant Biologists. All rights reserved.

separated from the other groups because of a local de-pression in the center of the mesophyll reflective oflarge intercellular airspaces. The model equation wasnot applied to this profile because of its divergencefrom the typical parabolic shape. Group (f) was uniquein that it contained profiles with chlorophyll maxima inthe adaxial domain (n = 7, R2 = 0.93, F(2,16) = 100.7, P,0.000), also from diverse lineages (i.e. fern, eudicot,monocot, magnoliid).

One-way ANOVAs were conducted to determine ifgroups of species with similar chlorophyll profiles(groups derived from the cluster analysis) differed inanatomical traits anticipated to influence chlorophylldistribution (i.e. leaf thickness, palisade fraction ofmesophyll, and vein depth). Groups (a) and (e) withonly one species were excluded from the analy-ses. Boxplots illustrating the differences in traits bygroup are shown in Figure 4. There was a statistically

Figure 3. Clustered display of chlorophyll profile data for 57 species. Dendrogram showing the relationship among specieschlorophyll profiles (A). The presence of six major profile schemes (groups a–f) is indicated. Mean chlorophyll profiles of thespecies in each cluster are shown in B. Chlorophyll content (as a percentage of the absolute chlorophyll maximum) is displayed asa function of relative depth (percentage depth from upper to lower edge of mesophyll). Points and error bars represent mean6 SE.SE bars less than ;3.0 are below resolution.

1410 Plant Physiol. Vol. 180, 2019

Borsuk and Brodersen

www.plantphysiol.orgon September 2, 2020 - Published by Downloaded from Copyright © 2019 American Society of Plant Biologists. All rights reserved.

significant difference in leaf thickness between groups[R2 = 0.25, F(3,54) = 5.65, P = 0.002] as determined byone-way ANOVA. A Tukey post-hoc test revealed thatleaves from the taxonomically diverse groups (c; meanleaf thickness = 0.2449 6 0.1386 mm, P = 0.002) and (d;mean leaf thickness = 0.2788 6 0.1002 mm, P = 0.020)were significantly thinner compared to leaves fromgroup (b), which was dominated by gymnosperms(mean leaf thickness = 0.4835 6 0.1780 mm). One-wayANOVA analysis indicated a statistically significantdifference in the palisade fraction of mesophyll be-tween groups [R2 = 0.23, F(3,47) = 4.38, P = 0.009]; aTukey post-hoc test showed that this difference wasonly significant (P = 0.004) between groups (b; meanpalisade fraction of mesophyll = 27.56%6 12.07%) and(c; mean palisade fraction of mesophyll = 46.81% 617.68%). There were no statistically significant differ-ences in vein depth between groups.

A General Model for Chlorophyll Distribution inLaminar Leaves

Quadratic regression was applied to the mean chlo-rophyll profile of the 57 species in our study, resultingin the following general model of relative chlorophylldistribution for the mean or typical leaf:

y ¼ 67:52þ 0:4149x  2 0:0029x2

(R2 = 0.94, F(2,16) = 127.3, p, 0.001; Fig. 5). Thus, forthe typical leaf from our dataset, chlorophyll increasedgradually as a function of leaf depth, with minor de-pressions in chlorophyll content at the outermost edgesof the mesophyll. The highest chlorophyll concentra-tion occurred at 83% (SE = 1.0) relative mesophyll depth,which is highly shifted towards the abaxial surface anddeeper than the mean vein position (;60% relative

mesophyll depth) and mean transition point betweenthe palisade and spongy tissue layers (;38% relativemesophyll depth). This extends the work of Johnsonet al. (2005) who first introduced a model equation(Eq. 1) to describe chlorophyll distribution in leaves oftwo species, R. catawbiense and A. fraseri.

Chlorophyll Distribution Models by Clade

Quadratic regression was applied to the mean chlo-rophyll profile of categorical sets of plants based onclade (fern, cycad, conifer, ginkgo, basal angiosperm,magnoliid, monocot, eudicot; Fig. 6, A–H). While me-sophyll depth was not a significant predictor of chlo-rophyll distribution for the clade-wise group of ferns(n = 3, R2 = 0.17, F(2,16) = 1.687, P = 0.216), it was a

Table 1. Numerical Coefficients for the Equation y = y0 + ax + bx2 forModels Based on Cluster Analysis, Clade, and Leaf Thickness

NA, Model equation could not be fitted to data.

Model y0 a b

Cluster a 10.50834 2.31279 20.01941Cluster b 50.45836 0.59300 20.00321Cluster c 76.51942 0.27804 20.00191Cluster d 58.27920 0.88874 20.00699Cluster e NA NA NACluster f 88.74845 20.38816 0.00250Fern NA NA NACycad 59.28052 0.53491 20.00348Conifer 45.26811 0.99436 20.00787Ginkgo 66.61461 0.70577 20.00672Basal angiosperm 61.51000 1.24990 0.00160Magnoliid 74.46951 0.30182 20.00223Monocot 65.78703 0.36470 20.00249Eudicot 71.05103 0.33471 20.00236Thin leaf 69.24298 0.35299 20.00229Thick leaf 64.36971 0.33783 20.00245

Figure 4. Differences in traits among groups with distinct chlorophyllprofiles. Box plots of leaf thickness (A), palisade fraction of mesophyll(B), and relative vein depth (C) by group derived from hierarchicalcluster analysis. Excludes groups a and e with single species. Boxesrepresent interquartile range and lines across boxes represent groupmedian. Whiskers extend from the upper and lower quartiles to thegroup maximum and minimum, respectively. Asterisks represent sam-ple outliers.

Plant Physiol. Vol. 180, 2019 1411

Spatial Distribution of Chlorophyll

www.plantphysiol.orgon September 2, 2020 - Published by Downloaded from Copyright © 2019 American Society of Plant Biologists. All rights reserved.

significant predictor and explained a high degree ofvariance in chlorophyll distribution for the sampledgroups of cycads (n = 9, R2 = 0.84, F(2,16) = 42.57, P ,0.001), conifers (n = 2, R2 = 0.96, F(2,16) = 211.7, P ,0.001), ginkgo (Ginkgo biloba; n = 1 species [mean of n = 6leaves and n = 3 profiles per leaf], R2 = 0.85,F(2,16) = 46.51, P , 0.001), basal angiosperm (n = 1species [mean of n = 6 leaves and n = 3 profiles perleaf], R2 = 0.96, F(2,16) = 170.2, P, 0.001), magnoliids(n = 6, R2 = 0.92, F(2,16) = 95.72, P, 0.001), monocots(n = 9, R2 = 0.69, F(2,16) = 17.47, P , 0.001), andeudicots (n = 26, R2 = 0.94, F(2,16) = 122, P , 0.001).Numerical coefficients for the regression curves areshown in Table 1.

Chlorophyll Distribution Models by Leaf Thickness

Quadratic regression was applied to the mean chlo-rophyll profile of categorical sets of plants based on leaf

thickness (Fig. 6, I and J). Mesophyll depth was a sig-nificant predictor of chlorophyll distribution for specieswith both categorically thin leaves (,0.5 mm; n = 47,R2 = 0.92, F(2,16) = 88.46, P , 0.001) and species withcategorically thick leaves ($0.5 mm; n = 10, R2 = 0.91,F(2,16) = 85.83, P , 0.001). Numerical coefficients forthe regression curves are shown in Table 1.

General Model Predictions versus Clade/Leaf ThicknessModel Predictions

Chlorophyll content predictions made by the cladeand leaf thickness model equations, which are signifi-cant for all but the fern model, correlated strongly withchlorophyll content predictions made by the general

Figure 5. General model of chlorophyll distribution. Mean chlorophylldistribution of 57 species as a function of mesophyll depth (A). Chlorophyllcontent (as a percentage of the absolute chlorophyll maximum) is displayedas a function of relative depth (percentage depth from upper to lower edgeof mesophyll). Quadratic regression is shown by the solid black curve.Points anderror bars representmean6 SE. Thedashed, vertical line indicatesthe relative depth of the transition between palisade and spongy mesophyllcell layers (38.64%6 1.60% SE.). The solid, vertical line indicates themeanvein depth relative to the adaxial edge of leaf mesophyll (59.76%6 1.27%SE.). Schematic of the typical leaf cross section parameterized with ap-proximate chlorophyll content according to the general model (B).

Figure 6. Chlorophyll distribution by clade and leaf thickness. Chlo-rophyll content (as a percentage of the absolute chlorophyll maximum)is displayed as a function of relative depth (percentage depth from upperto lower edge of mesophyll) for cladistic and leaf thickness subgroups:fern, n = 3 (A); cycad, n = 9 (B); conifer, n = 2 (C); ginkgo, n = 1 (D); basalangiosperm, n = 1 (E); magnoliid, n = 6 (F); monocot, n = 9 (G); eudicot,n = 26 (H); thin leaf, n = 47 (I); thick leaf, n= 10 (J). Regressions shown inthe form y ¼ y0 þ ax þ bx2 for the respective datasets. Points anderror bars represent mean 6 SE.

1412 Plant Physiol. Vol. 180, 2019

Borsuk and Brodersen

www.plantphysiol.orgon September 2, 2020 - Published by Downloaded from Copyright © 2019 American Society of Plant Biologists. All rights reserved.

model, or Equation 2. The general and subgroup modelpredictions were most highly correlated for the cycad,magnoliid, monocot, eudicot, thin leaf, and thick leafsubgroups (r = 0.99, P , 0.001), but also strongly cor-related for the conifer group (r = 0.97, P, 0.001), ginkgo(r = 0.67, P = 0.0018), and basal angiosperm (r = 0.87,P , 0.001).

Mesophyll Tissue Type and Vein Depth as Predictors ofChlorophyll Content Maxima

Given that the typical parabolic chlorophyll profile hasa distinguishable peak or global maximum, it was antic-ipated that this peak in chlorophyll content would beassociated with anatomical features of the leaf, i.e. at apredictable location relative to the transition from light-guiding palisade to light-diffusing spongy mesophyll orat a depth coincident with leaf veins. Indeed, by obser-vation, some chlorophyll profiles did peak at a similardepth as the veins (e.g. G. biloba; Supplemental Fig. S1).Yet, when linear regressionwas used to statistically assessthe strength of these anatomy-chlorophyll relationships,the spatial associations between anatomical traits andpeak chlorophyll content were very weak, and neithermesophyll tissue type transitions (r 20.05, n = 49,P = 0.710; Fig. 7A) nor vein depth (r = 0.08, n = 54,P = 0.956; Fig. 7B) were significant predictors of peakchlorophyll according to Pearson’s test.

Intraleaf Light Absorption and Chlorophyll Distribution

Light absorption profiles and chlorophyll profiles ofspecies from different clades (n = 5 species, Arengapinnata,Ulmus sp.,Dryopteris sp.,Coffea sp., and Boweniaspectabilis) were used to examine the relationship be-tween the intraleaf light environment and chlorophylldistribution. Monochromatic (blue, red, and green)light extinction profiles decayed exponentially throughthe leaf as seen in previous studies (e.g. Vogelmann and

Evans, 2002; Johnson et al., 2005; Brodersen andVogelmann, 2010). As anticipated from previous workby Vogelmann and Han (2000), red and blue light wereabsorbed strongly near the source of irradiation, whiledeeper portions of the mesophyll remained enriched ingreen light (Fig. 8). In an analysis of light absorptionprofiles generated from light entering opposite sides ofthe leaf, we found that the overlap of these profiles forred, green, and blue light occurred within a relativelynarrow region of the mesophyll, approximately be-tween the mean vein position and the position of thetransition between the palisade and spongy mesophyll.Chlorophyll content reached a maximum around thisnarrow band of tissue (Fig. 8).

DISCUSSION

Leaf Thickness, Palisade Fraction of Mesophyll, andIntercellular Airspaces May InfluenceChlorophyll Distribution

Although many of the leaves sampled showed chlo-rophyll distributions that were parabolic, with abax-ially shifted maxima, variations on and deviations fromthis trend were observed, indicating which anatomicaltraits may alter or influence chlorophyll allocation.Unsurprisingly, we found pronounced localized de-pressions in chlorophyll content in areas of leaves withhigh porosity, or large intercellular airspaces, such as inthe medial tissue of the aquatic plant E. crassipes (Fig. 2,G–I). Considering the results of the cluster analysis(Fig. 3B), chlorophyll content in group (b) increasedsharply across the photosynthetic domain, with a clearminimum in the upper leaf and maximum in the lowerleaf. This group was dominated by gymnosperms andhad statistically thicker leaves than plants in groups (c)and (d) and less palisade tissue than those in group (c).In contrast, group (c) had members with the thinnestleaves, was dominated by more derived taxa, and hadthe most uniform or flat chlorophyll distribution across

Figure 7. Spatial associations betweenanatomical traits and peak chlorophyllcontent. Mesophyll depth correspondingto peak chlorophyll concentration as afunction of the palisade to spongy meso-phyll transition (A) and vein depth (B).Depth of chlorophyll maxima measuredfrom species chlorophyll profiles as apercentage of mesophyll thickness. Veindepth measured from the adaxial edge ofleaf mesophyll as a percentage of meso-phyll thickness. Linear fit shown in black(statistically nonsignificant in both cases).

Plant Physiol. Vol. 180, 2019 1413

Spatial Distribution of Chlorophyll

www.plantphysiol.orgon September 2, 2020 - Published by Downloaded from Copyright © 2019 American Society of Plant Biologists. All rights reserved.

the photosynthetic domain. Taken together, this sug-gests that a thicker leaf with less developed palisadetissue (e.g. cycad) allocates less chlorophyll near theadaxial leaf surface, while a thinner leaf with moredeveloped palisade tissue (e.g. eudicot) allocates chlo-rophyll more uniformly throughout the chlorenchyma.Assuming that chlorophyll content increases to balancedecreases in available light (Bornman et al., 1991; Cuiet al., 1991), then the observed chlorophyll distributionswould occur because thinner leaves, particularly withhighly developed light-guiding palisade tissue, wouldexperience more of a homogenous internal light envi-ronment compared to a thick leaf, which would rapidlyattenuate light with increasing depth (Takahashi et al.,1994; Koizumi et al., 1998; Vogelmann and Han, 2000;Johnson et al., 2005; Brodersen and Vogelmann, 2010).Intraleaf gradients in CO2 may also influence the allo-cation of chlorophyll in bulk mesophyll domains. As-suming most laminar leaves have a higher relativeabundance of CO2 near stomata on the underside of theleaf, thick leaves might have a more pronounced CO2diffusion gradient and therefore less CO2 available forphotosynthesis in the upper mesophyll. We also notethat the localized depressions in chlorophyll content atboth adaxial and abaxial mesophyll edges observed inmost leaves could be due to limited mesophyll con-ductance in these areas (Earles et al., 2018). That is, if alarge proportion of the mesophyll cell wall has contactwith an epidermal cell, then the diffusion of CO2 wouldbe limited and chloroplasts within this region wouldunderperform. While beyond the scope of this work,future studies are anticipated to show the degree towhich variations in mesophyll porosity and/or con-nectivity influence CO2 availability, chlorophyll distri-bution, and chloroplast performance.

Common Patterns in Chlorophyll Distribution IndicateThat a General Chlorophyll Profile Scheme Is Conservedacross Phylogenetic Boundaries

When hierarchical cluster analysis was used toidentify common patterns in the chlorophyll profileswithout imposing a priori classification (Fig. 3), mostspecies (n = 48/57) were sorted into groups (b), (c), and(d), for which the mean chlorophyll profile convergedto a similar pattern; i.e. chlorophyll content increasedgradually as a function of mesophyll depth. Thus, ageneral pattern of chlorophyll distribution is exhibitedby plants as phylogenetically diverse as ferns, cycads,conifers, G. biloba, Illicium parviflorum (basal angio-sperm), magnoliids, monocots, and eudicots, repre-senting a diversity of growth habits, habitats, and leafmorphologies (e.g. tree, shrub, herbaceous, annual,perennial, evergreen, deciduous, etc.). Considering allspecies sampled in the study (n = 57), for the typicalspecies, chlorophyll content increased gradually as afunction of mesophyll depth from the adaxial surface,peaking deep within the leaf (;83% relative mesophylldepth). This is quantitatively described by a simplequadratic equation or general model (Eq. 2, R2 = 0.94,F(2,16) = 127.3, P , 0.001; Fig. 5).

Chlorophyll Content Maxima Are Well Positioned to TakeAdvantage of Relative Enrichment of Green Light inMedial and Abaxial Leaf Tissues and AbaxiallyIncident Irradiation

Our comparison of intraleaf light absorption profilesand chlorophyll profiles (Fig. 8) also suggests thatpeak chlorophyll content may respond not only to

Figure 8. Light absorption profiles andchlorophyll distribution in leaves. Subsetof taxa (n = 5 species, with n = 2–3 leavesmeasured per species): A. pinnata, B.spectabilis, Coffea sp., Ulmus sp., Dry-opteris sp. Points and error bars representspecies mean 6 SE. Leaves were irradiatedon their adaxial or abaxial surface withblue (squares), red (diamonds), or green(circles) light. Relative chlorophyll contentis displayed (open circles) as a function ofrelative depth (percentage depth from up-per to lower edge of mesophyll). Thedashed, vertical line represents the meanrelative depth of the transition betweenpalisade and spongy mesophyll (n = 3species, excludes B. spectabilis and Dry-poteris sp.; 47.73% 6 5.2% SE). The solid,vertical line represents mean vein depthrelative to the adaxial edge of leaf meso-phyll (n = 4 species, excludes Drypoterissp.; 62.3% 6 8.9% SE).

1414 Plant Physiol. Vol. 180, 2019

Borsuk and Brodersen

www.plantphysiol.orgon September 2, 2020 - Published by Downloaded from Copyright © 2019 American Society of Plant Biologists. All rights reserved.

irradiation incident on the upper leaf, but also to op-posing light gradients from both upper and lower leafsurfaces. We found that the red, green, and blue lightabsorption profiles overlapped within a relatively nar-row region of mesophyll that also coincided with thepositioning of the palisade to spongy transition, thelocation of the veins, and maximum chlorophyll con-tent. This region is also relatively enriched in green lightcompared with blue and red wavelengths, which isconsistent with the growing understanding of the roleof green light in driving photosynthesis within deepleaf tissue (Terashima et al., 2009; Johkan et al., 2012;Earles et al., 2017; Smith et al., 2017). While furtherstudies are needed, this could point to the optimizationof peak chlorophyll content where absorption of stronglight from the adaxial surface andweaker light from theabaxial surface is maximized.

Vein Depth and Mesophyll Tissue Type Are NotPredictive of Chlorophyll Content Maxima

We expected to find some coordination between thepeak chlorophyll content within the leaf and other an-atomical trait patterns that are believed to be importantfor optimizing photosynthesis. For example, the den-sity and spacing of veins within the leaf has been shownto have a remarkable scaling relationship across evo-lutionarily diverse species (Noblin et al., 2008), pre-sumably to most efficiently deliver water to the sites ofphotosynthesis and evaporation from the mesophyllsurface exposed to the intercellular airspace (Boyceet al., 2009). However, no statistical evidence for thecoordinated positioning of maximum chlorophyll con-tent with vein position was found (Fig. 7). Vein depthwas also not statistically different between groups ofspecies with distinctive chlorophyll profiles as identi-fied through cluster analysis, suggesting that vein lo-cation is not a driver of major patterns in chlorophylldistribution, at least in the 1D vertical profiles of bulkmesophyll measured in this study. Future studies areanticipated to show if there are additional patterns inchlorophyll distribution in 2- and 3-dimensions in theleaf, including around the vascular tissue in transverseand paradermal planes.

The General Model Predicts Bulk Mesophyll ChlorophyllDistribution in a Broad Range of Plant Lineages and forLeaves with Different Thicknesses

While we report that several variations in leaf formdo affect chlorophyll distribution, we also suggest thatthe general model (Eq. 2) is appropriate for describingchlorophyll distribution in most cases; this is becausechlorophyll distribution predictions based strictly oncategorical groupings of leaves (i.e. by clade and leafthickness; Fig. 6) were highly correlated with thosegiven by the general model (r = 0.99, P , 0.001 for

cycad, magnoliid, monocot, eudicot, thin leaf, and thickleaf groups).

CONCLUSION

While further studies are warranted to determine themechanistic underpinnings of the chlorophyll profileschemes characterized in this study, the main trendshared by the majority of species—as well as the devi-ations from this trend in thick leaves—suggests thatfundamental physical constraints, such as intraleafgradients in light, are central to chlorophyll distributionin laminar leaves. It is worth noting that the leaves inthis study were all relatively thin (,1 mm) and mostexhibited C3 photosynthesis. Variation in chlorophylldistribution could occur outside these phenotypicparameters; for example, the C4 grass Zea mays(Supplemental Fig. S1) shows a depression in chloro-phyll content near the midpoint of the depth profile,potentially due to the highly modified Kranz anatomyof the bundle sheath that occupies a significant portionof the leaf cross sectional area. For our datasets wheremesophyll depth failed to predict chlorophyll distri-bution (e.g. clade-wise for ferns), this could in futurestudies be retested by increasing the sample sizeor taxonomic representatives. Expanding the epi-illumination chlorophyll fluorescence dataset, particu-larly within a genus or species and in two and threedimensions, is an opportunity for further research. Forexample, the observed similarity between chlorophyllprofiles of congeneric pairs points to the possibilityof nuanced modification of chlorophyll distribu-tion according to shared traits. The response of thesetraits and of the spatial allocation of chlorophyll withinthe leaf to different environmental conditions awaitstesting.In summary, this study provides empirical chloro-

phyll distribution data and model equations for manyecologically and commercially relevant species, quan-titatively defines a common chlorophyll distributionscheme across phylogenetic and functional groupboundaries, and initiates an understanding of themechanisms underlying chlorophyll distribution inbulk mesophyll domains. Taken together, these find-ings advance our understanding of first principles inintraleaf physiology and move us toward more accu-rate in silico photosynthesis modeling and a clearerrelationship between chlorophyll distribution, thelight environment within a leaf, and photosyntheticperformance.

MATERIALS AND METHODS

Plant Material

Mature, fully expanded leaves were collected from various environments(Marsh Botanical Garden, New Haven, CT; Yale Marsh Greenhouse, NewHaven, CT; Montgomery Botanical Center, Coral Gables, FL; University ofVermont Campus, Burlington, VT; University of Vermont Greenhouse,

Plant Physiol. Vol. 180, 2019 1415

Spatial Distribution of Chlorophyll

www.plantphysiol.orgon September 2, 2020 - Published by Downloaded from Copyright © 2019 American Society of Plant Biologists. All rights reserved.

Burlington,VT; private garden,Monterey,CA; naturalizedhabitat,NewHaven,CT; Supplemental Table S1). Spinacia oleracea and Arabidopsis (Arabidopsisthaliana) wild type were grown in commercial and controlled growth chamberenvironments, respectively. In outdoor environments, leaves were collectedfrom sun-exposed, south-facing sides of the plant canopy. Highly contrastingleaf phenotypes (i.e. sun or shade) were avoided where distinctive. All specieshad laminar or flattened leaf forms. Excised leaveswere adapted to darkness fora minimum of 20 min before being prepared for chlorophyll fluorescencemeasurements. This period of dark adaptation was implemented as a controlfor possible chloroplast movement within cells as a response to the externallight environment (Trojan and Gabrys, 1996).

Anatomical Measurements

Leaf thickness, upper epidermis thickness (including hypodermis wherepresent), lower epidermis thickness, mesophyll thickness, percentage of me-sophyll cross sectional area occupied by palisade mesophyll, and vein depth(measured from the adaxial edge of leafmesophyll to the center of the vein)weremeasured from images of leaf cross sections. All measurements were replicatedtwiceper leaf on2 to6 leaves for each species. In three cases (Begonia erythrophylla,Dryopteris sp., Zamia fairchildiana), veins were excluded from or not discernablein the chlorophyll fluorescence images. Mean and SE were calculated for allmeasurements. Species were categorized as having thick or thin leaves sub-jectively based on the nonuniform distribution of sampled leaf thickness fre-quencies by 0.05 mm bins (thin , 0.50 mm, thick $ 0.50 mm).

Measurement of Relative Chlorophyll Distributionwithin Leaves

Relative chlorophyll distribution within leaves was measured using epi-illumination fluorescence microscopy similar to the method of Vogelmannand Evans (2002). Transverse hand sections cut several cells thick weremounted on lightly wetted filter paper on top of a glass slide, such that thesample did not dry out during imaging and while preventing the infiltration ofintercellular airspaces with mounting liquid (water). The sample was thenplaced on the stage of a microscope (Olympus BX43, Olympus America) andirradiated with epi-illumination at 490 nm generated from a broad-spectrumLED light source (Lumen 300-LED, Prior Scientific Instruments) after passingthrough a filter. Light emitted via chlorophyll fluorescence passed through abarrier filter (680 nm, half bandwidth = 16 nm, S10-680F; Corion Filters) andwas imaged with a digital Peltier-cooled CCD camera (PIXIS 1024B, PrincetonInstruments) using shutter times of 70 to 150 ms. Individual profiles weremeasured from the images by drawing lines from the adaxial edge of the me-sophyll to the abaxial edge of themesophyll with a line width of 50 to 100 pixels,overwhich themean intensitywas given; in general 100-pixel width (equivalentto ;60 mm at 203 magnification or ;120 mm at 103 magnification) was used;however, thinner widths were used as necessary when nonphotosyntheticfeatures such as veins were spaced less than 100 pixels apart. Measurementsexcluded epidermal cells, veins, and other conspicuous nonphotosyntheticstructures such as hypodermis. Chlorophyll and depth data for each profilewere normalized by dividing all values by the global absolute chlorophyll ordepth maxima, respectively. The mean of three profiles was measured for eachleaf sample, and the mean of 2 to 6 leaf samples was measured per species.Estimates of absolute fluorescence intensity among different samples were notpossible. This was due to variability in exposure needed to image differentlysized samples at appropriate focal lengths, and additionally the temporalvariation in signal due to Kautsky decay, i.e. the decline in fluorescence in-tensity as a function of exposure time under continuous irradiation. As dis-cussed in Vogelmann and Han (2000), Kautsky decay changes the relativeintensity of the fluorescence signal but does not alter the relative distribution offluorescence imaged from the sample. The entire epi-illumination process for asingle sample was completed within approximately 30 s. Differential detectionof chlorophyll a and b by the epi-illumination method was considered as asource of bias in the shape of the measured chlorophyll profiles, yet is assumedto be minimal due to precedent by Vogelmann and Evans (2002). These authorsfound a strong linear relationship for in vivo paradermal chlorophyll concen-tration as a function of fluorescence detected by the epi-illumination method.Additionally, excitation and emission bands used here were well within peaksfor detection of both chlorophyll a and b. Finally, our protocol for normalizingthe fluorescence signal for each sample removes potential bias in chlorophylla/b ratios across species.

Cluster Analysis of Species Chlorophyll Profiles

Hierarchical cluster analysis was used to group species chlorophyll profilesaccording to similar distribution schemes. Euclidean distances between chlo-rophyll intensity data points were calculated at 21 loci (0%, 5%, 10%, . 100%relative mesophyll depth) and clusters were agglomerated using completelinkage. Analysis was performed using the R statistical program (R Core Team,2018). Visual assessment of clustering patterns indicated the presence of sixmajor groups. The small subcluster containing Begonia bowerae andAntirrhinummajus could have been treated as a seventh group; however, to avoid over-splitting, and consistent with observed patterns in the species chlorophyllprofiles, this was grouped with its neighboring major cluster. To aid in com-parison of major patterns in chlorophyll distribution between groups, a meanmodel was generated from the individual chlorophyll profiles of species in eachgroup. A regression of the form y = y0+ ax + bx2was applied to each groupmeanmodel and tested for significance and fit (Johnson et al., 2005). Domains wherethe upper and lower epidermal cell walls met the chlorenchyma cell walls(0%–5% and 95%–100% relative depth, respectively) were excluded from theregression. These domains represent a nonphotosynthetic transition point be-tween tissue types, which is retained in the data as a marker of the transitionpoint in tissue morphology.

General Model of Chlorophyll Distribution

A mean model approach was used to develop a general equation for chlo-rophyll distribution in laminar leaves. The mean model, or mean specieschlorophyll profile, was calculated by averaging relative chlorophyll intensityvalues at common relative depth values for all 57 species in the study. Mean SE

was calculated to evaluate the variation in chlorophyll intensity among speciesas a function ofmesophyll depth. A regression curve of the form y = y0 + ax + bx2

was applied to the meanmodel and tested for fit (Johnson et al., 2005). Domainswhere the upper and lower epidermal cell walls met the chlorenchyma cellwalls (0%–5% and 95%–100% relative depth, respectively) were excluded fromthe regression. These domains represent a nonphotosynthetic transition pointbetween tissue types, which is retained in the data as a marker of the transitionpoint in tissue morphology.

Measurement of Light Absorption Profiles in Leaves

Chlorophyll fluorescencewas also used to estimate the relative absorption ofred, green, and blue light through leaf samples using previously reportedmethods (Takahashi et al., 1994; Koizumi et al., 1998; Vogelmann and Han,2000; Vogelmann and Evans, 2002; Brodersen and Vogelmann, 2010). Freshleaf sections were cut to;1 cm2with a razor blade andmounted in a 3D-printedsample holder on the stage of amicroscope (Olympus BX43, Olympus America)with the transverse edge normal to the objective. The abaxial or adaxial side ofthe leaveswere irradiatedwithmonochromatic red (660 nm), green (532 nm), orblue (488 nm) light generated by one of three lasers (red solid-state laser, model#BWN-660-10E, BandW Tek; green solid-state laser, model #DY20B, PowerTechnology; blue solid-state laser, model LRS-473-TM-10-5, LaserGlow Tech-nologies). The laser light was attenuated with neutral density filters such thatthe light incident on the leaf surface was of equal intensity (1200 mmol m22s21)measured by a LI-190 (Li-Cor) light meter mounted in the same position as theleaf sample holder. As the leaf was illuminated with the monochromatic laserlight, the transverse (orthogonal) surface of the leaf was observed for fluores-cence emitted from the leaf tissue with a Peltier-cooled CCD camera (PIXIS1024B, Princeton Instruments). The laser spot (1-mm radius) was directedonto the 1-cm2 leaf surface centered on the cut surface being observed. Con-sequently, the observed fluorescence would only arise from chloroplasts up to1 mm from the cut transverse surface. Light profiles were measured from theresulting digital images where light absorbance was proportional (1:1) to theimaged fluorescence, or pixel intensity (Vogelmann and Evans, 2002). Mea-surements were taken from the adaxial edge of the mesophyll to the abaxialedge of the mesophyll and excluded conspicuous nonphotosynthetic structuressuch as veins. Themean of three profiles wasmeasured for each leaf sample andthe mean of 2 to 3 leaf samples was measured per species.

All images were processed using the FIJI (Schindelin et al., 2012) distributionof ImageJ (Rueden et al., 2017) software. All statistical calculations andgraphing were performed using R (R Core Team, 2018), with the exception ofANOVA and Tukey analyses, which were performed using Minitab 18Statistical Software (2017).

1416 Plant Physiol. Vol. 180, 2019

Borsuk and Brodersen

www.plantphysiol.orgon September 2, 2020 - Published by Downloaded from Copyright © 2019 American Society of Plant Biologists. All rights reserved.

Supplemental Data

The following supplemental materials are available.

Supplemental Table S1. The 57 plant species studied and associated traits.

Supplemental Figure S1. Chlorophyll distribution as a function of meso-phyll depth for 57 species.

ACKNOWLEDGMENTS

The authors thank Dr. Jay Wason (University of Maine) and Dr. AdamRoddy (Yale University) for discussions, Dr. Jonathan Reuning-Scherer (YaleUniversity) for statistical consultation, and Dr. Joshua Gendron (Yale Univer-sity), theMarsh Botanical Garden (NewHaven, CT) andMontgomery BotanicalCenter (Coral Gables, FL) for plant materials. The authors declare that there isno conflict of interest.

Received January 24, 2019; accepted March 19, 2019; published April 3, 2019.

LITERATURE CITED

Aalto T, Juurola E (2002) A three-dimensional model of CO2 transport inairspaces and mesophyll cells of a silver birch leaf. Plant Cell Environ 25:1399–1409

Adams III WW, Terashima I (2018) The Leaf: A Platform for PerformingPhotosynthesis, Vol 44. Springer, Dordrecht, the Netherlands

Bornman JF, Vogelmann TC, Martin G (1991) Measurement of chlorophyllfluorescence within leaves using a fibreoptic microprobe. Plant CellEnviron 14: 719–725

Boyce CK, Brodribb TJ, Feild TS, Zwieniecki MA (2009) Angiosperm leafvein evolution was physiologically and environmentally transformative.Proc Biol Sci 276: 1771–1776

Brodersen CR, Vogelmann TC (2010) Do changes in light direction affectabsorption profiles in leaves? Funct Plant Biol 37: 403–412

Cui M, Vogelmann TC, Smith WK (1991) Chlorophyll and light gradients insun and shade leaves of Spinacia oleracea. Plant Cell Environ 14: 493–500

Earles JM, Théroux-Rancourt G, Gilbert ME, McElrone AJ, Brodersen CR(2017) Excess diffuse light absorption in upper mesophyll limits CO2

drawdown and depresses photosynthesis. Plant Physiol 174: 1082–1096Earles JM, Theroux-Rancourt G, Roddy AB, Gilbert ME, McElrone AJ,

Brodersen CR (2018) Beyond porosity: 3D leaf intercellular airspacetraits that impact mesophyll conductance. Plant Physiol 178: 148–162

Earles JM, Buckley TN, Brodersen CR, Busch FA, Cano FJ, Choat B, EvansJR, Farquhar GD, Harwood R, Huynh M, et al (2019) Embracing 3Dcomplexity in leaf carbon-water exchange. Trends Plant Sci 24: 15–24

Evans JR, Vogelmann TC (2003) Profiles of 14C fixation through spinachleaves in relation to light absorption and photosynthetic capacity. PlantCell Environ 26: 547–560

Evans JR, Vogelmann TC (2006) Photosynthesis within isobilateral Euca-lyptus pauciflora leaves. New Phytol 171: 771–782

Ho QT, Berghuijs HNC, Watté R, Verboven P, Herremans E, Yin X, RettaMA, Aernouts B, Saeys W, Helfen L, et al (2016) Three-dimensionalmicroscale modelling of CO2 transport and light propagation in tomatoleaves enlightens photosynthesis. Plant Cell Environ 39: 50–61

Ichiro T, Hiroki O, Takashi F, Riichi O (2016) Light environment within aleaf. II. Progress in the past one-third century. J Plant Res 129: 353–363

Johkan M, Shoji K, Goto F, Hahida S, Yoshihara T (2012) Effect of greenlight wavelength and intensity on photomorphogenesis and photosyn-thesis in Lactuca sativa. Environ Exp Bot 75: 128–133

Johnson DM, Smith WK, Vogelmann TC, Brodersen CR (2005) Leaf ar-chitecture and direction of incident light influence mesophyll fluores-cence profiles. Am J Bot 92: 1425–1431

Koizumi M, Takahashi K, Mineuchi K, Nakamura T, Kano H (1998) Lightgradients and the transverse distribution of chlorophyll fluorescence inmangrove and camellia leaves. Ann Bot 81: 527–533

Minitab 18 Statistical Software (2017) Minitab, Inc., State College, PA.www.minitab.com

Morison JIL, Lawson T, Cornic G (2007) Lateral CO2 diffusion inside di-cotyledonous leaves can be substantial: Quantification in different lightintensities. Plant Physiol 145: 680–690

Nishio JN, Sun J, Vogelmann TC (1993) Carbon fixation gradients acrossspinach leaves do not follow internal light gradients. Plant Cell 5: 953–961

Noblin X, Mahadevan L, Coomaraswamy IA, Weitz DA, Holbrook NM,Zwieniecki MA (2008) Optimal vein density in artificial and real leaves.Proc Natl Acad Sci USA 105: 9140–9144

Oguchi R, Douwstra P, Fujita T, Chow WS, Terashima I (2011) Intra-leafgradients of photoinhibition induced by different color lights: implica-tions for the dual mechanisms of photoinhibition and for the applicationof conventional chlorophyll fluorometers. New Phytol 191: 146–159

Ort DR, Merchant SS, Alric J, Barkan A, Blankenship RE, Bock R, CroceR, Hanson MR, Hibberd JM, Long SP, et al (2015) Redesigning pho-tosynthesis to sustainably meet global food and bioenergy demand. ProcNatl Acad Sci USA 112: 8529–8536

Palta JP (1990) Leaf chlorophyll content. Remote Sens Rev 5: 207–213Pieruschka R, Schurr U, Jensen M, Wolff WF, Jahnke S (2006) Lateral

diffusion of CO2 from shaded to illuminated leaf parts affects photo-synthesis inside homobaric leaves. New Phytol 169: 779–787

Porra RJ, Thompson WA, Kriedemann PE (1989) Determination of accu-rate extinciton coefficients and simultaneous equations for assayingchlorophylls a and b extracted with four different solvents: Verificationof the concentration of chlorophyll standards by atomic absorptionspectroscopy. Biochim Biophys Acta Bioenerg 975: 384–394

R Core Team (2018) R: A language and environment for statistical com-puting. R Foundation, Vienna, Austria. https://www.r-project.org/

Reich PB (2014) The world-wide ‘fast-slow’ plant economics spectrum: Atraits manifesto. J Ecol 102: 275–301

Rueden CT, Schindelin J, Hiner MC, DeZonia BE, Walter AE, Arena ET,Eliceiri KW (2017) ImageJ2: ImageJ for the next generation of scientificimage data. BMC Bioinformatics 18: 529

Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T,Preibisch S, Rueden C, Saalfeld S, Schmid B, et al (2012) Fiji: An open-source platform for biological-image analysis. Nat Methods 9: 676–682

Shipley B, Lechowicz MJ, Wright I, Reich PB (2006) Fundamental trade-offsgenerating the worldwide leaf economics spectrum. Ecology 87: 535–541

Slattery RA, Grennan AK, Sivaguru M, Sozzani R, Ort DR (2016) Lightsheet microscopy reveals more gradual light attenuation in light-greenversus dark-green soybean leaves. J Exp Bot 67: 4697–4709

Smith HL, McAusland L, Murchie EH (2017) Don’t ignore the green light:Exploring diverse roles in plant processes. J Exp Bot 68: 2099–2110

Smith WK, Vogelmann TC, Delucia EH, Bell DT, Shepherd KA (1997)Leaf form and photosynthesis: Do leaf structure and orientation interactto regulate internal light and carbon dioxide? BioScience 47: 785–793

Sun J, Nishio JN, Vogelmann TC (1998) Green light drives CO2 fixationdeep within leaves. Plant Cell Physiol 39: 1020–1026

Takahashi K, Mineuchi K, Nakamura T, Koizumi M, Kano H (1994) Asystem for imaging transverse distribution of scattered light and chlo-rophyll fluorescence in intact rice leaves. Plant Cell Environ 17: 105–110

Terashima I, Fujita T, Inoue T, Chow WS, Oguchi R (2009) Green lightdrives leaf photosynthesis more efficiently than red light in strong whitelight: revisiting the enigmatic question of why leaves are green. PlantCell Physiol 50: 684–697

Terashima I, Hanba YT, Tholen D, Niinemets Ü (2011) Leaf functionalanatomy in relation to photosynthesis. Plant Physiol 155: 108–116

Tholen D, Zhu X-G (2011) The mechanistic basis of internal conductance: Atheoretical analysis of mesophyll cell photosynthesis and CO2 diffusion.Plant Physiol 156: 90–105

Tholen D, Boom C, Zhu X-G (2012) Opinion: Prospects for improvingphotosynthesis by altering leaf anatomy. Plant Sci 197: 92–101

Tholen D, Éthier G, Genty B (2014) Mesophyll conductance with a twist.Plant Cell Environ 37: 2456–2458

Trojan A, Gabrys H (1996) Chloroplast distribution in Arabidopsis thaliana (L.)depends on light conditions during growth. Plant Physiol 111: 419–425

Vogelmann TC, Evans JR (2002) Profiles of light absorption and chloro-phyll within spinach leaves from chlorophyll fluorescence. Plant CellEnviron 25: 1313–1323

Vogelmann TC, Gorton HL (2014) Leaf: Light Capture in the Photosyn-thetic Organ. Springer, Dordrecht, the Netherlands, pp 363–377

Vogelmann TC, Han T (2000) Measurement of gradients of absorbed lightin spinach leaves from chlorophyll fluorescence profiles. Plant Cell En-viron 23: 1303–1311

Zwieniecki MA, Boyce CK (2014) Evolution of a unique anatomical pre-cision in angiosperm leaf venation lifts constraints on vascular plant e-cology. Proc Biol Sci 281: 20132829

Plant Physiol. Vol. 180, 2019 1417

Spatial Distribution of Chlorophyll

www.plantphysiol.orgon September 2, 2020 - Published by Downloaded from Copyright © 2019 American Society of Plant Biologists. All rights reserved.